Introduction To Nastran Sol 200 Size Optimization

Introduction to Nastran SOL 200
Design Sensitivity and Optimization

PRESENTED BY: CHRISTIAN APARICIO
Nastran SOL 200 questions? Email me: christian@ the-engineering-lab.com

Motivation
It is my intent that after you read this
guide, you will be one step closer
towards performing a unique
procedure only a limited number of
engineers can do.
That is, optimizing structures

automatically with Nastran SOL 200.
Kind Regards,
Christian Aparicio Shape Optimization is used to find optimum boundaries of this pulley
Nastran SOL 200 questions? Email me: christian@ the-engineering-lab.com 2

Motivation Slide 2
What is optimization? Slide 4
What is the optimization problem statement? Slide 13
Table of What is size optimization? Slide 16
Contents Optimization Examples Slide 21
How to set up Nastran SOL 200 Slide 31
Nastran SOL 200 Learning Resources Slide 39
Final Comments Slide 40
Bonus Section What is Sensitivity Analysis? Slide 41

What is optimization?

1 , 2 = 12 + 22
Suppose we define and plot the

function below:
1 , 2 = 12 + 22

1 , 2 = 12 + 22
Initial Point
1, 2 = (3, 4
We start at point: 1 , 2 = (3, 4
The objective is to find the minima

of 1, 2

1 , 2 = 12 + 22
The optimizer in Nastran SOL 200 finds

the minima at (0, 0).
The optimum or minima is

found at (0, 0)

1 , 2 = 12 + 22
Optimization is the process of finding

minima or maxima of functions.
Optimum

1 , 2 = 12 + 22
Lets continue the previous example

and add this constraint: 2 2
1 1 , 2 = 12 + 1
2
2 2
1 1 , 2 = 12 + 1
2

1 , 2 = 12 + 22
The feasible region is the set of

design variables 1 , 2 that satisfy
all the constraints (In this example
the feasible region is everything
What does the optimizer do in this
outside of the ellipsoid)
scenario? See next slide.
The infeasible region is the region

where any combination of design
variables 1 , 2 will violate one or
more constraints (In this example,
the infeasible region is inside of the
ellipsoid)

1 , 2 = 12 + 22
The optimizer will move towards the

global minima

Once it reaches the constraint, the
optimizer will move along the
constraint until the minima is found
Optimum

What is the optimization
problem statement?

This is the optimization problem
statement for the last example
The optimization problem statement is

composed of 3 items:
1. Design Variables What
parameters are allowed to vary?
2. Design Objective What quantity

is to be minimized or maximized?
3. Design Constraints What

quantities are constrained?

1 , 2 = 12 + 22
It is important to always have the

optimization problem statement
drafted before performing an
optimization

What is size
optimization?

1D Elements 2D Elements
Consider the example finite element

model to the right.
Disassembled View
On first glance, the structure does not

seem too significant.
Assembled View

After assigning parameters such as
thickness to the 2D elements and
beam cross section dimensions to the
1D elements, the structures looks as
shown on the right.

t
Size optimization is the process of

setting parameters as design variables. DIM2
For example, the thickness and the DIM1
dimension 1 and 2 can be set as design
variables.
1: t, thickness
2: DIM1
3: DIM2 DIM4 DIM3 DIM4
Other parameters that can be set as
design variables include: Youngs
modulus or density of a material. The
complete list of parameters that can
be set as design variables is extensive.

Size Optimization
Note that many optimization types

exist. To the right are some of the
many types.
Topology Optimization Shape Optimization

The remainder of this guide will only
apply to size optimization.
Topometry Optimization Topography Optimization

Optimization Examples
with Nastran SOL 200

Example 1 - Optimization of a composite
laminate
A cylindrical tube is composed
of a composite laminate with
the following layup: [85/- Edge is constrained from translating
85/60/-60/60/-60/85/-85]. in the z direction -85
Each ply has a thickness of .01 85
inches. Internal pressure of -60
400 psi 60
Loading and Constraints -60
60
-85
The tube is allowed to freely 85
expand radially outward. An
internal pressure of 400 psi is
applied.
Click here to
watch this tutorial

laminate
Optimization Problem Statement Optimization Results
Design Variables: and 90 degrees Initial Design Optimized Design
Thickness Variables Laminas 1, 2, 7, 8 should have Mass: 1.60 lbfs2/in Mass: .3386 lbfs2/in
the same orientation angle,
x1, x2, x3, x4, x5, x6, x7, x8 but with opposing signs Layup: [85/-85/60/-60/60/- Layup: [45/-45/0/0/0/0/45/-45]
correspond to the thickness 60/85/-85]
Laminas 3, 4, 5, 6 should have Ply Thicknesses: .0021 in.
of lamina 1 through 8,
respectively the same orientation angle, Ply Thicknesses: .0100 in. Max strength ratio: .9
but with opposing signs
Each thickness variable shall Max strength ratio: 21.0
be equal to x1 The orientation angles are
only allowed to be in 5
x1 is allowed to range degree increments, e.g. 90,
between .001 and 10. 85, 80 -80, -85, -90
Orientation Variables Design Objective
x9, x10, x15, x16 correspond
Minimize the weight
to the outer layer pair angles,
i.e. 85 degree plies Design Constraints
x11, x12, x13, x14 correspond The strength ratio for each
to the core pair angles, i.e. 45 lamina shall be less than .9
degree plies
The orientation angles are
allowed to vary between -90.

laminate
Objective Value Design Variable History of Lamina Orientation Angles
1.80 100
1.60 80
60
1.40
Orientation Angle [degrees]

40
1.20
Weight [lbfs2/in]
20 X9
1.00 X11
0
X10
0.80
-20
X12
0.60 -40 X13
X14
0.40 -60
X15
0.20 -80
X16
-100
0.00
14
15
10
11
12
13
16
17
Initial Design
1
2
3
4
5
6
7
8
9
Design Cycle Design Cycle

Example 2 - Model Matching / System
Identification / Correlation to Experiment
A beam, fixed on one end, has a circular Experimental Results
cross section with a radius of 2 in. The Mode 1 (First Bending Mode)
Mode 1 Disp. Mode 3 Disp.
length of the beam is 30 in. Experiment
Node Component Value Component Value 1.2
has revealed the expected mode shapes
Z Displacement (Relative)
for modes 1 and 3. A modes analysis of 3 z or 3 direction 0.0143 x or 1 direction 0.1204 1
the finite element (FE) models shows a 6 z or 3 direction 0.1741 x or 1 direction 0.5431 0.8
discrepancy between the FE model and 9 z or 3 direction 0.6381 x or 1 direction 0.9216 0.6
experiment. 0.4
Initial Design
Experimental Results
0.2
Optimization is used to find a radius of
0
the cross section that will produce FE
results comparable to experiment. Click here to 1 2 3 4 5 6 7 8 9 10 11
Node
watch this tutorial
Fixed Mode 3 (First Extensional Mode)
1.2
X Displacement (Relative)
1
0.8
0.6
Initial Design
0.4
Length: 30 in Radius: 2in Experimental Results
0.2
E = 1 * 107 psi 0
1 2 3 4 5 6 7 8 9 10 11
v = .3
Node
Density = .01 lbf * s2 / in4

Design Variables: Design Constraint The following plots show the values for the mode shapes of the initial design
x1: The radius of 3 elements is allowed to vary The RSS of mode 3 is to be less than .002 and optimized design. Note that the optimized or final design has normal mode
between .1 and 10. inches. g1 = RSS = sqrt ( d2 + e2 + f2 )
shapes that align with experiment
Design Objective d = (.1204 r501)
Mode 1 (First Bending Mode)
e = (.5431 r502)
Experimental data is available regarding the 1 st
mode shape f = (.9216 r503) 1.2
Z Displacement (Relative)
1
The objective is to minimize the root sum of r501, r502, r503 are the x displacements at nodes
3, 6, 9, respectively, for mode 3 0.8
squares (RSS) between the experiment and FE
0.6 Initial Design
results
0.4 Final Design
f = RSS = sqrt ( a2 + b2 + c2 )
0.2
a = (.0143 r801) Experimental Results
0
b = (.1741 r802) 1 2 3 4 5 6 7 8 9 10 11
c = (.6381 r803) Node
r801, r802, r803 are the z displacements at nodes Mode 3 (First Extensional Mode)
3, 6, 9, respectively, for mode 1
1.2
X Displacement (Relative)
1
0.8
0.6 Initial Design
0.4 Final Design

0.2 Experimental Results
0
1 2 3 4 5 6 7 8 9 10 11
Node

Objective Value Design Variable History
0.25 4.5
4.0
0.20
3.5
Root Sum of Squares (RSS)
3.0
0.15
Radius [in]
2.5
2.0 X1
0.10
1.5
1.0
0.05
0.5
0.00 0.0
Initial Design 1 2 3 4 Initial Design 1 2 3 4
Design Cycle Design Cycle

Example 3 Buckling Optimization of a
Thin Walled Cylinder
A thin walled cylinder reinforced structure while ensuring stress is
with ring stiffeners is subjected not exceeded and buckling does
to an axial compressive load. not occur. This optimization is an
The initial design is far from example of multi-discipline
exceeding the allowable stresses optimization since the structure
in the stiffeners and wall. The is optimized for both static and
buckling factor is well above 1, buckling analyses.
so buckling will not occur.
Optimization is used to vary two DIM1: 1 in.

structural dimension, wall and DIM2: .03 in
stiffener thickness, so as to DIM3: 1 in.
minimize the mass of the DIM4: .5 in
Click here to
watch this tutorial

Design Variables: Design Constraints Initial Design Optimized Design
x1: The thickness of the thin wall is Statics Subcase Mass: 5.76 lbf Mass: 1.77 lbf
allowed to vary. Bounds: .001 < x1 < x1 = Thickness of wall = .03 in x1 = Thickness of wall = .0135 in
10. The maximum beam stress in the
stiffener is allowed to be no x2 = Thickness of stiffener = .03 in x2 = Thickness of stiffener = .0016 in
x2: The thickness of the stiffener greater than 25000.
cross section is allowed to vary. Buckling factor = 3.6 Buckling factor = .99 1.0
Bounds: .001 < x2 < 10. The maximum von Mises stress Stiffener and shell stresses within limits
for the z1 and z2 fibers of the Stiffener and shell stresses within
Design Objective thin wall shall be no greater than
25000.
limits
Minimize Weight
Buckling Subcase
The buckling load factor shall be
no less than 1.0
x2
x1 Buckling Shape (B.F. = 3.6) Buckling Shape (B.F. = .99)

Objective Value
7.00
6.00
Weight [lbf]
5.00
4.00
3.00
2.00
1.00
0.00
Initial 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27
Design
Design Cycle
Design Variable History

0.035
Thickness [in.]
0.030
0.025
0.020
0.015
0.010
0.005
0.000
Initial 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27
Design
Design Cycle X1, Thickness of Wall
X2, Thickness of Stiffener

How to Set Up Nastran
SOL 200
A STEP-BY-STEP PROCEDURE FOR CONVERTING . BDF OR .DAT FILES
TO SOL 200

SOL 10x
BDF
To perform optimization, your original

.bdf or .dat file must be converted to
SOL 200
The Nastran SOL 200 Web App

facilitates this process
Nastran SOL 200 Web App
Click here to
watch a demo on SOL 200
BDF

Click here to
watch this tutorial
Three-Bar Truss Optimization Example
This example can be found in the MSC Nastran Design Sensitivity and Optimization Users Guide. The text
and images below have been extracted from that users guide.
A common task in design optimization is to reduce

the mass of a structure subjected to several load
conditions. Figure 8-1 shows a simple three-bar truss
that must be built to withstand two separate loading
conditions. Note that these two loads subject the
outer truss members to both compressive as well as
tensile loads. Due to the loading symmetry, we
Step 0 Draft the optimization problem expect the design to be symmetric as well. As an
statement exercise, we'll show how to enforce this symmetry
Design Variables There are 3 design using design variable linking. - MSC Nastran Design
variables Sensitivity and Optimization Users Guide, Chapter 8:
x1: Area of element 1 Example Problems, Three-Bar Truss
x2: Area of element 2
x3: Area of element 3
The variables are allowed to vary
between .001 and 100.
The area of element 3 must equal to the
area of element 1
x3 = x1
Design Objective
Minimize weight
Design Constraints
g1 - The x and y displacement at node 4
is allowed to be within -.2 and .2 inches
g2 - The axial stress in each element is
allowed to be within -15000 psi and
20000 psi
Both of these constraints are applied to
Subcase 1 and 2

1
Step 1 Create Design Variables
1. Copy the text found in the .bdf file and

paste it in the web app 6
3
2. Click on Identify Designable Properties
The app will identify parameters in your
model that can be set as design variables
3. Click the plus ( + ) icons of the parameters

to set the parameters as design variables
4. Adjust the lower and upper bounds of the

design variables
5. Create the link between design variables

x3 and x1 4
6. Note that as you use the Web App, the
necessary SOL 200 bulk data entries are
automatically created

1
Step 2 Create Design Objective
1. Select Weight as the objective

from the list of available
responses
2. Set the objective to minimize

2

1
Step 3 Create Design Constraints
1. Click on the plus icons next to

Displacement and Stress to create
two constraints 2
2. Configure the Displacement and

Stress constraints
3. Create a constraint group 3
4. Apply the constraint group to all

the subcases 4

Step 4 Export
1. Click on Download BDF to

download new .bdf files to your
desktop 1 1
2. Start MSC Nastran, select
model.bdf file to perform the
optimization
2
1

A Patran is used to view structural results. Left: Original
structural results of the initial design. Right: New
structural results of the optimized design. B The .csv file is opened in excel.
Next Steps
A. The structural results of the

optimized design can be viewed in
a pre/post processor
B. The change of objective and

design variables during the
optimization can be viewed in
excel
C. The original .bdf file can be

updated with new optimized
entries found in the .pch file
C Left: The .pch file has the updated PROD entries. Right:
The original .bdf file is updated with optimized PROD
entries.

Resource Link
Nastran SOL 200 Nastran SOL 200 Optimization Tutorials on YouTube

Link
Learning Hours of Nastran SOL 200 tutorials are available on my YouTube channel.
Resources Nastran SOL 200 Web App
This web application will enable you to convert you existing .bdf files and convert
Email me for access
Here is a quick summary of helpful them to SOL 200.
tutorials and guides for Nastran SOL
200. Free Live Training
Link
Attend a live training course instructed by me and over 7 hours in length.
Guidance from an Optimization Expert

As always, if you have a Nastran SOL
200 question, you are welcome to If you are working an optimization project of your own and would like support or
Email me
email me. guidance, you are welcome to email me.
MSC Nastran Design Sensitivity and Optimization Users Guide
Link
This guide is where I gained a majority of my optimization knowledge and I highly
recommend it as a reference.

One last optimization example
The objective is to find the maximum of , = followed the path of steepest accent (highest
2 2 contained within the ellipsoid. x and y gradient), but led to 3 different maxima. When
are allowed to vary. Three optimizations were finding optimums of functions with higher
performed, each represented by the colors purple, dimensions, keep in mind that multiple optimums
red and green. Each optimization had a different may exist and that each optimum will depend on
initial point. In each scenario, the optimizer your initial design variables.
Optima
Final Comments
You are now one step closer to
optimizing structures automatically.
I am here to support you, and if you
have any questions regarding Nastran
SOL 200, you are welcome to email
me.
Thank you for reading this guide and

stay motivated. Initial Points
Sincerely, , = 2 2
Christian Aparicio

Bonus Section
What is Sensitivity Analysis?

Sensitvity analysis the process of calculating
partial derivatives.
The partial derivatives are commonly

referred to as sensitivity coefficients or just
sensitivities when using Nastran SOL 200.
Since sensitivities are nothing more than

rates of change, design variables can be
selected based on their influence on
outputs. A question such as this can be
answered, which design variable, x1 or x2, if
varied, will best minimize the function f?
Another example, if the goal is to minimize a Image source: khanacademy.org Introduction to Partial Derivatives
weight function, it would be logical to select
design variables that have sensitivities to the
weight.

Let us consider a simple example as
shown to the right.

Partial Derivatives / Sensitivities
The sensitivities are computed by

hand and will be compared to the
sensitivities computed by Nastran SOL
200.

This is the sensitivity output by
Nastran SOL 200 and can be found in
the .f06 file.
Note that Nastran SOL 200 produces

sensitivities in line with the hand
calculated sensitivities.

Absolute Sensitivities
=6
1
There are different types of

sensitivities you may have to consider. Normalized =6 3
Sensitivities 1 1
= 18
To the right are some sensitivities and
how they are related.
Normalized Relative 1 3
=6
Sensitivities 1 (1 , 2 25
= .72

Introduction To Nastran Sol 200 Size Optimization

Diunggah oleh

Informasi Dokumen

Judul Asli

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

Introduction To Nastran Sol 200 Size Optimization

Diunggah oleh

Hak Cipta:

Format Tersedia

Introduction to Nastran SOL 200

Design Sensitivity and Optimization

Nastran SOL 200 questions? Email me: christian@ the-engineering-lab.com

That is, optimizing structures

Nastran SOL 200 questions? Email me: christian@ the-engineering-lab.com 2

Nastran SOL 200 questions? Email me: christian@ the-engineering-lab.com 3

Nastran SOL 200 questions? Email me: christian@ the-engineering-lab.com 4

Suppose we define and plot the

Nastran SOL 200 questions? Email me: christian@ the-engineering-lab.com 5

The objective is to find the minima

Nastran SOL 200 questions? Email me: christian@ the-engineering-lab.com 6

The optimizer in Nastran SOL 200 finds

The optimum or minima is

Nastran SOL 200 questions? Email me: christian@ the-engineering-lab.com 7

Optimization is the process of finding

Nastran SOL 200 questions? Email me: christian@ the-engineering-lab.com 8

Lets continue the previous example

Nastran SOL 200 questions? Email me: christian@ the-engineering-lab.com 9

The feasible region is the set of

The infeasible region is the region

Nastran SOL 200 questions? Email me: christian@ the-engineering-lab.com 10

The optimizer will move towards the

Nastran SOL 200 questions? Email me: christian@ the-engineering-lab.com 11

Nastran SOL 200 questions? Email me: christian@ the-engineering-lab.com 12

Nastran SOL 200 questions? Email me: christian@ the-engineering-lab.com 13

The optimization problem statement is

2. Design Objective What quantity

3. Design Constraints What

Nastran SOL 200 questions? Email me: christian@ the-engineering-lab.com 14

It is important to always have the

Nastran SOL 200 questions? Email me: christian@ the-engineering-lab.com 15

Nastran SOL 200 questions? Email me: christian@ the-engineering-lab.com 16

Consider the example finite element

On first glance, the structure does not

Nastran SOL 200 questions? Email me: christian@ the-engineering-lab.com 17

Nastran SOL 200 questions? Email me: christian@ the-engineering-lab.com 18

Size optimization is the process of

Nastran SOL 200 questions? Email me: christian@ the-engineering-lab.com 19

Note that many optimization types

Topology Optimization Shape Optimization

Topometry Optimization Topography Optimization

Nastran SOL 200 questions? Email me: christian@ the-engineering-lab.com 20

Nastran SOL 200 questions? Email me: christian@ the-engineering-lab.com 21

Nastran SOL 200 questions? Email me: christian@ the-engineering-lab.com 22

Nastran SOL 200 questions? Email me: christian@ the-engineering-lab.com 23

Orientation Angle [degrees]

Nastran SOL 200 questions? Email me: christian@ the-engineering-lab.com 24

Nastran SOL 200 questions? Email me: christian@ the-engineering-lab.com 25

c = (.6381 r803) Node

0.4 Final Design

Nastran SOL 200 questions? Email me: christian@ the-engineering-lab.com 26

Nastran SOL 200 questions? Email me: christian@ the-engineering-lab.com 27

Optimization is used to vary two DIM1: 1 in.

Nastran SOL 200 questions? Email me: christian@ the-engineering-lab.com 28

x1 Buckling Shape (B.F. = 3.6) Buckling Shape (B.F. = .99)

Nastran SOL 200 questions? Email me: christian@ the-engineering-lab.com 29

Design Variable History

Nastran SOL 200 questions? Email me: christian@ the-engineering-lab.com 30

Nastran SOL 200 questions? Email me: christian@ the-engineering-lab.com 31

To perform optimization, your original

The Nastran SOL 200 Web App

Nastran SOL 200 Web App

Nastran SOL 200 questions? Email me: christian@ the-engineering-lab.com 32